Tcc tradutor mobile - fei

André Antonelli David Machado James do Carmo Orientador: Prof. Dr. Paulo Sérgio Rodrigues

Tradutor Mobile com Segmentação Baseada em Entropia

Monograa apresentada ao Curso de Graduação em Ciência da Computação do Centro Universitário da FEI, como requisito parcial para a obtenção do grau de Bacharel em Ciência da Computação.

São Bernardo do Campo, SP Junho de 2011

Abstract
This paper presents a methodology for translating texts, extracted from an image previously captured by the camera of a mobile device. In most cases, the implementation of a image processing system, is much more complex because of the limited hardware resources of a mobile device. The methodology presented here is based on the entropy concept, constantly used by researchers in various elds of application. Using the camera of the device, the image is captured, then a module based on the entropy concept, segments the image in order to adequately separate the characters in the scene. The regions of these characters are passed to a module for the optical character recognition, which are extracted from the image. Afterwards, these characters are submitted to a online Web API, responsible for the translation. Using the area previously dened by the user, the recognition of the region where the text is will be possible. With that information, it is possible to change the perspective vision, and the translated text can be overlaid on the original image. This system will be able to assist travelers who are in other countries or simply stimulate the learning of new languages.

Keywords:

Translation, Entropy, Segmentation, Optical Character Recognition.

i

Resumo
Este trabalho apresenta uma metodologia para tradução de textos, extraídos de uma imagem previamente capturada pela câmera de um dispositivo móvel. Na maioria das vezes, dispositivos deste tipo por possuírem recursos de hardware limitados, tornam mais complexa a implementação de um sistema de processamento de imagens. A